Studying Memory Retrieval at the Dynamic Neural Network Level

Massachusetts General Hospital, Boston MA

Investigators

Linked publications & trials

Abstract

? DESCRIPTION (provided by applicant): A fundamental question in neuroscience is how memories are represented by the collective pattern of spiking within neural populations. A related question is whether the same collective pattern used to represent memories during learning are reactivated during later recall, or if the nature of the population code is vitally different. While prior theoretical work and more recent work in animal models have provided valuable initial insights into how memories are likely represented, the neural process by which memories are actually retrieved at the network-wide level, what exact aspects of the first-, second- and higher-order network structure are informative of the retrieved memories, and by what rapid sub-second dynamic do such informative patterns evolve within individual trials remain fundamentally unknown. This limited understanding is particularly true of hetero-associative memory processes such as cued recollection in which the items being recalled are both absent and completely unique from the presented items used to cue their retrieval. Here, we aim to systematically define, for the first time, the combined network-level processes that underlie these basic forms of auto-associative (recognition) and hetero-associative (recollection) memory. Towards these ends, we will use the shared expertise of the two principal investigators to perform simultaneous multi-electrode recordings from frontal and temporal cortical populations in Rhesus macaques; devise and test novel analysis methodologies that can both reliably infer the collective first-, second- and higher-order functional network structures from stochastic spiking data; track their rapid sub-second dynamics within individual trials; identify which structures are informative of the memories being recalled; determine when and to what extent spiking network structures observed during learning reactivate during recall within individual-day sessions; and, perhaps most importantly, determine whether pattern reactivation, at the spiking network-level, is causatively related to recall accuracy at the behavioral level. Th present proposal will allow us to directly test, for the first time, a number of central hypotheses on memory processing using a novel set of technical and methodological innovations that will have broad practical implications to the study of memory-related developmental, behavioral and neurodegenerative disorders.